summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/r200/r200_span.c
diff options
context:
space:
mode:
Diffstat (limited to 'src/mesa/drivers/dri/r200/r200_span.c')
-rw-r--r--src/mesa/drivers/dri/r200/r200_span.c307
1 files changed, 0 insertions, 307 deletions
diff --git a/src/mesa/drivers/dri/r200/r200_span.c b/src/mesa/drivers/dri/r200/r200_span.c
deleted file mode 100644
index 9783678028..0000000000
--- a/src/mesa/drivers/dri/r200/r200_span.c
+++ /dev/null
@@ -1,307 +0,0 @@
-/*
-Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- * Keith Whitwell <keith@tungstengraphics.com>
- */
-
-#include "main/glheader.h"
-#include "main/imports.h"
-#include "main/colormac.h"
-#include "swrast/swrast.h"
-
-#include "r200_context.h"
-#include "r200_ioctl.h"
-#include "r200_state.h"
-#include "r200_span.h"
-#include "r200_tex.h"
-
-#define DBG 0
-
-/*
- * Note that all information needed to access pixels in a renderbuffer
- * should be obtained through the gl_renderbuffer parameter, not per-context
- * information.
- */
-#define LOCAL_VARS \
- driRenderbuffer *drb = (driRenderbuffer *) rb; \
- const __DRIdrawablePrivate *dPriv = drb->dPriv; \
- const GLuint bottom = dPriv->h - 1; \
- GLubyte *buf = (GLubyte *) drb->flippedData \
- + (dPriv->y * drb->flippedPitch + dPriv->x) * drb->cpp; \
- GLuint p; \
- (void) p;
-
-#define LOCAL_DEPTH_VARS \
- driRenderbuffer *drb = (driRenderbuffer *) rb; \
- const __DRIdrawablePrivate *dPriv = drb->dPriv; \
- const GLuint bottom = dPriv->h - 1; \
- GLuint xo = dPriv->x; \
- GLuint yo = dPriv->y; \
- GLubyte *buf = (GLubyte *) drb->Base.Data;
-
-#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS
-
-#define Y_FLIP(Y) (bottom - (Y))
-
-#define HW_LOCK()
-
-#define HW_UNLOCK()
-
-
-
-/* ================================================================
- * Color buffer
- */
-
-/* 16 bit, RGB565 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_RGB
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
-
-#define TAG(x) r200##x##_RGB565
-#define TAG2(x,y) r200##x##_RGB565##y
-#define GET_PTR(X,Y) (buf + ((Y) * drb->flippedPitch + (X)) * 2)
-#include "spantmp2.h"
-
-/* 32 bit, ARGB8888 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x) r200##x##_ARGB8888
-#define TAG2(x,y) r200##x##_ARGB8888##y
-#define GET_PTR(X,Y) (buf + ((Y) * drb->flippedPitch + (X)) * 4)
-#include "spantmp2.h"
-
-
-/* ================================================================
- * Depth buffer
- */
-
-/* The Radeon family has depth tiling on all the time, so we have to convert
- * the x,y coordinates into the memory bus address (mba) in the same
- * manner as the engine. In each case, the linear block address (ba)
- * is calculated, and then wired with x and y to produce the final
- * memory address.
- * The chip will do address translation on its own if the surface registers
- * are set up correctly. It is not quite enough to get it working with hyperz too...
- */
-
-/* extract bit 'b' of x, result is zero or one */
-#define BIT(x,b) ((x & (1<<b))>>b)
-
-static GLuint
-r200_mba_z32( driRenderbuffer *drb, GLint x, GLint y )
-{
- GLuint pitch = drb->pitch;
- if (drb->depthHasSurface) {
- return 4 * (x + y * pitch);
- }
- else {
- GLuint b = ((y & 0x7FF) >> 4) * ((pitch & 0xFFF) >> 5) + ((x & 0x7FF) >> 5);
- GLuint a =
- (BIT(x,0) << 2) |
- (BIT(y,0) << 3) |
- (BIT(x,1) << 4) |
- (BIT(y,1) << 5) |
- (BIT(x,3) << 6) |
- (BIT(x,4) << 7) |
- (BIT(x,2) << 8) |
- (BIT(y,2) << 9) |
- (BIT(y,3) << 10) |
- (((pitch & 0x20) ? (b & 0x01) : ((b & 0x01) ^ (BIT(y,4)))) << 11) |
- ((b >> 1) << 12);
- return a;
- }
-}
-
-static GLuint
-r200_mba_z16( driRenderbuffer *drb, GLint x, GLint y )
-{
- GLuint pitch = drb->pitch;
- if (drb->depthHasSurface) {
- return 2 * (x + y * pitch);
- }
- else {
- GLuint b = ((y & 0x7FF) >> 4) * ((pitch & 0xFFF) >> 6) + ((x & 0x7FF) >> 6);
- GLuint a =
- (BIT(x,0) << 1) |
- (BIT(y,0) << 2) |
- (BIT(x,1) << 3) |
- (BIT(y,1) << 4) |
- (BIT(x,2) << 5) |
- (BIT(x,4) << 6) |
- (BIT(x,5) << 7) |
- (BIT(x,3) << 8) |
- (BIT(y,2) << 9) |
- (BIT(y,3) << 10) |
- (((pitch & 0x40) ? (b & 0x01) : ((b & 0x01) ^ (BIT(y,4)))) << 11) |
- ((b >> 1) << 12);
- return a;
- }
-}
-
-
-/* 16-bit depth buffer functions
- */
-#define VALUE_TYPE GLushort
-
-#define WRITE_DEPTH( _x, _y, d ) \
- *(GLushort *)(buf + r200_mba_z16( drb, _x + xo, _y + yo )) = d;
-
-#define READ_DEPTH( d, _x, _y ) \
- d = *(GLushort *)(buf + r200_mba_z16( drb, _x + xo, _y + yo ));
-
-#define TAG(x) r200##x##_z16
-#include "depthtmp.h"
-
-
-/* 24 bit depth, 8 bit stencil depthbuffer functions
- */
-#define VALUE_TYPE GLuint
-
-#define WRITE_DEPTH( _x, _y, d ) \
-do { \
- GLuint offset = r200_mba_z32( drb, _x + xo, _y + yo ); \
- GLuint tmp = *(GLuint *)(buf + offset); \
- tmp &= 0xff000000; \
- tmp |= ((d) & 0x00ffffff); \
- *(GLuint *)(buf + offset) = tmp; \
-} while (0)
-
-#define READ_DEPTH( d, _x, _y ) \
- d = *(GLuint *)(buf + r200_mba_z32( drb, _x + xo, \
- _y + yo )) & 0x00ffffff;
-
-#define TAG(x) r200##x##_z24_s8
-#include "depthtmp.h"
-
-
-/* ================================================================
- * Stencil buffer
- */
-
-/* 24 bit depth, 8 bit stencil depthbuffer functions
- */
-#define WRITE_STENCIL( _x, _y, d ) \
-do { \
- GLuint offset = r200_mba_z32( drb, _x + xo, _y + yo ); \
- GLuint tmp = *(GLuint *)(buf + offset); \
- tmp &= 0x00ffffff; \
- tmp |= (((d) & 0xff) << 24); \
- *(GLuint *)(buf + offset) = tmp; \
-} while (0)
-
-#define READ_STENCIL( d, _x, _y ) \
-do { \
- GLuint offset = r200_mba_z32( drb, _x + xo, _y + yo ); \
- GLuint tmp = *(GLuint *)(buf + offset); \
- tmp &= 0xff000000; \
- d = tmp >> 24; \
-} while (0)
-
-#define TAG(x) r200##x##_z24_s8
-#include "stenciltmp.h"
-
-
-/* Move locking out to get reasonable span performance (10x better
- * than doing this in HW_LOCK above). WaitForIdle() is the main
- * culprit.
- */
-
-static void r200SpanRenderStart( GLcontext *ctx )
-{
- r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
- R200_FIREVERTICES( rmesa );
- LOCK_HARDWARE( rmesa );
- r200WaitForIdleLocked( rmesa );
-
- /* Read & rewrite the first pixel in the frame buffer. This should
- * be a noop, right? In fact without this conform fails as reading
- * from the framebuffer sometimes produces old results -- the
- * on-card read cache gets mixed up and doesn't notice that the
- * framebuffer has been updated.
- *
- * In the worst case this is buggy too as p might get the wrong
- * value first time, so really need a hidden pixel somewhere for this.
- */
- {
- int p;
- driRenderbuffer *drb =
- (driRenderbuffer *) ctx->WinSysDrawBuffer->_ColorDrawBuffers[0];
- volatile int *buf =
- (volatile int *)(rmesa->dri.screen->pFB + drb->offset);
- p = *buf;
- *buf = p;
- }
-}
-
-static void r200SpanRenderFinish( GLcontext *ctx )
-{
- r200ContextPtr rmesa = R200_CONTEXT( ctx );
- _swrast_flush( ctx );
- UNLOCK_HARDWARE( rmesa );
-}
-
-void r200InitSpanFuncs( GLcontext *ctx )
-{
- struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
- swdd->SpanRenderStart = r200SpanRenderStart;
- swdd->SpanRenderFinish = r200SpanRenderFinish;
-}
-
-
-
-/**
- * Plug in the Get/Put routines for the given driRenderbuffer.
- */
-void
-radeonSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
-{
- if (drb->Base.InternalFormat == GL_RGBA) {
- if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
- r200InitPointers_RGB565(&drb->Base);
- }
- else {
- r200InitPointers_ARGB8888(&drb->Base);
- }
- }
- else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
- r200InitDepthPointers_z16(&drb->Base);
- }
- else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
- r200InitDepthPointers_z24_s8(&drb->Base);
- }
- else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
- r200InitStencilPointers_z24_s8(&drb->Base);
- }
-}